[Question]: did internal reranker model support nvidia gpu ?

[chminsc]

Describe your problem

for example, netease bce-reranker-base model

[chminsc]

or maybe use ollama to serve reranker model would be a better practice?

[KevinHuSh]

Ollama is better.

[chminsc]

Ollama is better.

ollama is still full of bugs working together with ragflow. for example, can't add reranker to ragflow through ollama ( I did see your suggustion to add by xinference, but not working for gguf models)
embedding is not woking for long text files(lots of bug reporting here)